摘要
提出了基于差影法粗分割与多模板匹配的人脸检测方法。首先利用差影法分割出可能包含人脸的区域;然后使用双眼模板进行粗筛选,进一步减小可能人脸的搜索范围;最后利用人脸模板和灰度规则进行匹配和验证。此方法具有较强的鲁棒性和可扩展性,减少了运算量,降低了误检率,提高了检测系统的综合性能。
A human face detection method based on coarse segmentation of difference image and multi-template matching is proposed. Firstly, the region which may contain faces is segmented out by using of difference image; secondly, the template of two eyes is used for coarse selection to minish the area of searching faces; lastly, face templates and grey rule are used to match and verify. This method not only possesses higher robustness and expansion, but also reduces calculated capacity, decreases error detection ratio and increases synthesis capability of detection system.
出处
《计算机工程与设计》
CSCD
2004年第10期1648-1650,共3页
Computer Engineering and Design
基金
国家863高技术计划基金(2001AA144170)
